Bearing Monitoring

Bearing Fault Frequencies: BPFO, BPFI, BSF, and FTF

The four bearing defect frequencies computed from rolling element bearing geometry and shaft speed, with the formulas, worked examples, and diagnostic significance of each.

Key Principle

Bearing defect frequencies are not approximations. They are exact mathematical results from the geometry of the bearing. A bearing with N rolling elements of diameter d running at shaft frequency f has an outer race defect frequency of exactly (N/2) × f × (1 − d/D × cosθ). Knowing this frequency precisely is what makes vibration-based bearing condition monitoring possible.

Frequency → Amplitude noise FTF (cage) ~0.4× shaft BSF (ball) ~2.4× shaft BPFO outer race 1× shaft BPFI inner race ±1× sidebands Outer race: BPFO Inner race: BPFI ± 1× Ball: BSF Cage: FTF FTF BSF BPFO 1× BPFI Typical bearing defect frequency positions (not to scale). BPFO and BPFI are the most commonly detected. BPFI sidebands appear because the inner race rotates with the shaft.

What Bearing Fault Frequencies Represent

Each of the four fault frequencies corresponds to a specific physical event inside the bearing. BPFO is the rate at which rolling elements pass over a fixed point on the outer race. BPFI is the rate at which rolling elements pass over a fixed point on the rotating inner race. BSF is the spinning rate of an individual rolling element about its own axis. FTF (also called cage frequency) is the orbital rate of the cage around the bearing axis.

When a defect forms on a race or rolling element, each time a rolling element contacts that defect it generates a mechanical impact. These impacts repeat at the corresponding fault frequency. In a vibration spectrum, energy concentrated at the fault frequency and at its integer harmonics indicates a defect at that specific component.

The practical consequence: if you know the shaft speed and the exact bearing geometry, you know exactly where in the frequency spectrum to look for a defect on each bearing component. There is no need to scan blindly.

Bearing Geometry Variables

The same four variables appear in every fault frequency formula. Define them clearly before computing anything.

Variable Description Typical units
N Total number of rolling elements (balls or rollers) in one row Count (integer)
d Ball or roller diameter mm or inches
D Pitch diameter: the diameter of the circle passing through the centers of all rolling elements. Also called bearing mean diameter or ball pass diameter. mm or inches (same as d)
θ Contact angle: the angle between the line of ball-to-race contact and the radial plane of the bearing. Purely radial ball bearings: 0°. Angular contact bearings: typically 15°, 25°, or 40°. Radial roller bearings: 0°. Degrees (convert to radians for calculation)
f Shaft rotational frequency. f = RPM / 60 Hz

How to find these values: the bearing manufacturer's catalog or technical datasheet lists d, D, N, and θ for every bearing by designation number. SKF, NSK, Timken, and Schaeffler all publish searchable online catalogs with complete geometry data. Do not estimate these values from physical measurement. Small errors in d or D propagate into errors in all four fault frequencies, which shifts the frequencies you are monitoring away from the true defect frequency.

BPFO: Ball Pass Frequency Outer Race

The outer race is stationary in most bearing configurations. Rolling elements orbit around it at the cage frequency. BPFO is the number of times rolling elements strike a given point on the outer race per second.

BPFO = (N/2) × f × (1 − (d/D) × cos(θ)) N = number of rolling elements f = shaft speed in Hz (RPM / 60) d = ball/roller diameter D = pitch diameter θ = contact angle (convert degrees to radians: radians = degrees × π / 180)

Worked example: SKF 6205

Bearing parameters from the SKF catalog: N = 9 balls, d = 7.938 mm, D = 38.5 mm, θ = 0°. Shaft speed = 1,750 RPM, so f = 1750 / 60 = 29.17 Hz.

cos(0°) = 1, so the geometry term simplifies to (1 − 7.938/38.5) = (1 − 0.2062) = 0.7938.

BPFO = (9/2) × 29.17 × 0.7938 = 4.5 × 29.17 × 0.7938 = 104.2 Hz

BPFO is always less than (N/2) × f because the geometry term (1 − d/D × cosθ) is always less than 1 for any practical bearing geometry. An outer race defect on this bearing at 1,750 RPM produces impacts at 104.2 Hz and at harmonics 208.4 Hz, 312.6 Hz, and so on.

BPFI: Ball Pass Frequency Inner Race

The inner race rotates with the shaft. BPFI is the rate at which rolling elements contact a fixed point on the rotating inner race, accounting for both the orbital motion of the cage and the shaft rotation. Because the inner race moves toward oncoming rolling elements, the contact rate is higher than for the outer race.

BPFI = (N/2) × f × (1 + (d/D) × cos(θ))

Worked example: SKF 6205 (continued)

Using the same parameters: geometry term = (1 + 0.2062) = 1.2062.

BPFI = 4.5 × 29.17 × 1.2062 = 158.3 Hz

BPFI is always greater than (N/2) × f. A useful cross-check: BPFO + BPFI = N × f. In this example: 104.2 + 158.3 = 262.5, and N × f = 9 × 29.17 = 262.5. If your values do not satisfy this relationship, recheck the inputs.

Inner race defects develop sidebands spaced at shaft frequency (BPFI ± 1×f, BPFI ± 2×f) because the defect rotates through the loaded and unloaded zones of the bearing once per shaft revolution, which amplitude-modulates the impact train at shaft frequency.

BSF: Ball Spin Frequency

BSF is the spinning rate of each individual rolling element about its own axis as it rolls along the inner and outer race surfaces.

BSF = (D / (2 × d)) × f × (1 − ((d/D) × cos(θ))²)

Worked example: SKF 6205 (continued)

D/(2d) = 38.5 / (2 × 7.938) = 38.5 / 15.876 = 2.425. The squared term: (0.2062)² = 0.04252. So: (1 − 0.04252) = 0.9575.

BSF = 2.425 × 29.17 × 0.9575 = 67.7 Hz

BSF is typically sub-synchronous, meaning it falls below shaft speed for most radial ball bearing geometries. Ball defects appear at BSF and at 2×BSF in the spectrum. The 2×BSF harmonic is often more prominent because a defect on the surface of a ball contacts both the inner race and the outer race once per revolution of the ball, producing two impacts per ball revolution rather than one.

FTF: Fundamental Train Frequency (Cage Frequency)

FTF is the orbital rate of the cage around the bearing axis. The cage keeps rolling elements equally spaced as they orbit, so FTF describes how fast the entire set of rolling elements travels around the shaft.

FTF = (f / 2) × (1 − (d/D) × cos(θ))

Worked example: SKF 6205 (continued)

FTF = (29.17 / 2) × 0.7938 = 14.585 × 0.7938 = 11.58 Hz

FTF is always significantly below shaft frequency. For most common radial ball bearings it falls between 0.30 and 0.45 times shaft speed. At 1,750 RPM on this bearing, the cage orbits at 11.58 Hz, which is 0.397 times shaft speed.

FTF peaks appear in the very low frequency portion of the vibration spectrum. They also appear as a modulation frequency on the BPFO and BPFI peaks: when a cage is damaged or worn, BPFO sidebands appear spaced at FTF. This modulation pattern is a secondary indicator of cage distress rather than a race defect.

Summary: All Four Frequencies for SKF 6205 at 1,750 RPM

Frequency Formula term Result (Hz) Relationship to shaft speed
BPFO (N/2) × f × (1 − d/D) 104.2 Hz 3.57× shaft
BPFI (N/2) × f × (1 + d/D) 158.3 Hz 5.43× shaft
BSF (D/2d) × f × (1 − (d/D)²) 67.7 Hz 2.32× shaft
FTF (f/2) × (1 − d/D) 11.6 Hz 0.40× shaft

Cross-check: BPFO + BPFI = 104.2 + 158.3 = 262.5. N × f = 9 × 29.17 = 262.5. Confirmed.

Harmonics and Sideband Patterns

Fault frequencies appear not only at their fundamental but also at integer harmonics: 2×BPFO, 3×BPFO, and so on. The presence and relative amplitude of harmonics indicate fault severity.

Early-stage defects typically show a prominent fundamental with little harmonic energy. As the defect grows, energy distributes across multiple harmonics. Late-stage defects may show broad harmonic families and elevated broadband noise floor in addition to discrete fault frequency peaks.

BPFI peaks develop sidebands spaced at shaft frequency (BPFI ± 1×f, BPFI ± 2×f) due to load-zone modulation. As the inner race defect rotates through the loaded zone (where rolling elements are pressed against the defect under load) and then through the unloaded zone, the amplitude of the impact changes cyclically at shaft frequency. This modulation appears as sidebands in the spectrum.

BPFO peaks generally do not develop shaft-frequency sidebands because the outer race is stationary and the defect location does not move relative to the load zone.

Use the Bearing Fault Frequency Calculator to compute all four frequencies for your specific bearing.

Open Calculator

Obtaining Bearing Geometry Data

The accuracy of all four fault frequencies depends entirely on the accuracy of the geometry inputs. There are three steps to getting correct values.

Step 1: Identify the bearing designation. The designation is stamped on the bearing outer ring and is recorded on the motor nameplate, in the equipment maintenance record, or in the OEM documentation. If the designation is not accessible without disassembly, check the equipment BOM or contact the motor manufacturer.

Step 2: Look up the bearing in the manufacturer's catalog. SKF, NSK, Timken, and Schaeffler all publish searchable online catalogs with complete geometry data including d, D, N, and θ for every bearing in their range. Search by designation number and locate the technical data table.

Step 3: Use the exact catalog values. Do not round d or D to a convenient number. Do not estimate D from the inner or outer ring diameter. Small errors compound: a 2% error in d/D shifts all four frequencies by roughly 0.4% to 0.8%, which can displace the expected fault frequency outside the frequency resolution of the analysis, causing a real defect to appear as broadband noise rather than a discrete peak.

Automated Bearing Fault Frequency Monitoring

Tractian automatically computes BPFO, BPFI, BSF, and FTF from the bearing designation for each connected asset and monitors amplitude at each frequency continuously. Alerts fire when energy at a fault frequency crosses configurable thresholds.

See Mechanical Monitoring
mechanicalmonitoring.com is an educational reference resource published by Tractian. Formulas and examples are provided for general reference. Always verify bearing geometry data from the official manufacturer catalog for your specific bearing designation before use in a monitoring program.